Diagnostic Genes Identification and Molecular Classification Patterns Based on Oxidative Stress-Related Genes in Ischemic Stroke

Ischemic Stroke (IS) has been ever-threatening to public health. Oxidative stress participated in the IS process and induced neuron death, lacking pathological elucidation and clinical diagnosis. Acquiring data from Gene Expression Omnibus (GEO) and Gene Set Enrichment Analysis (GSEA) database, we screened differently expressed genes and selected candidate diagnosis-related genes via random forest and SVM-RFE method. Subsequently we established a corresponding diagnosis model and eventually conducted the immune infiltration analysis. Parallelly, we conducted the consensus clustering and performed weighted gene co-expression network analysis (WGCNA) to figure out co-expressed gene modules in clusters respectively. Then we conducted functional enrichment for featured pathways of each cluster, and presented the KEGG pathways on the proteomic level. Eventually, the single-cell analysis was conducted to further explore the oxidative stress in IS, including pseudo-time analysis and cell-cell communication. Through machine learning algorithms, 9 key hub genes were selected and a diagnosis prediction model was conducted. The immune infiltration analysis revealed correlations between immune cells, inflammatory factors and differently expressed genes. The clusters obtained by consensus clustering indicated different functional pathways. The single-cell analysis revealed that endothelial cells were obviously affected by oxidative stress in IS. Associations between oxidative stress-induced and IS was clarified. New perspectives were provided for IS pathological elucidation, diagnosis and treatment usage.